Child pages
  • UMA legal subgroup notes

Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

2019-12-03

Attending: Eve, Domenico, Lisa, Nancy (regrets: Tim)

Lisa's and Eve's "Beyond Consent" article is now in a near-final version; Lisa kindly agreed (when asked by Eve) to distribute this version to the UMA WG.

Consensus: We looked at the Typical Use Case section of the doc and concluded that we do like having the "typical use case" for technical roles presented, followed by the formal definitions of said roles. Let's also plan to publish the spreadsheet as a companion Report because it can serve as the "source of truth" of definitions and relationships. We can prepare an Annex or Appendix or whatever we want to call it in this Report that consists of tables, along with a citation to the spreadsheet that has "live" data. We can say "In case of any discrepancy, the spreadsheet has the most accurate data" etc.

Assumption: We note here – but don't want to note in the Report, because it would complicate things for the readers – that EHR data typically gets selectively copied to a portal, and doesn't "live" there. If we elucidated all of our assumptions and caveats in the "typical use case", we'd never get anywhere. Nancy is clarifying for our group exactly how SMEs would split hairs on primary storage of health data. We agree that our audience will be motivated simply to understand the UMA paradigm at a relatively high level at this stage.

Lisa asks (paraphrased): When we say the resource owner "manages" resources at the RS and "controls" access at the AS, is she doing that at an app of some sort? Why are we saying Alice is interacting directly with services here and not apps? She does use apps, but we don't talk about them in protocol-related conversations because those interactions aren't "on the wire"; there are no UMA-dictated messages involved. However, those who deploy UMA-enabled systems do have to concern themselves with the implementation and deployment of such apps! Eve's recent webinar with Steve Giovannetti covered the four elements of integration and deployment needed:

  • Deploy an AS. We have noted many times that policy engine functionality is something each AS can compete on because it lets ROs get more sophistication in policy-setting ability. However, often in healthcare use case conversations, the idea of policy interoperability comes up too.
  • Integrate/create your client app (if you already have a client, you have to UMA-enable it) – is there room for creating "UMAlet" middleware? Gluu has talked about its open-source middleware in the past.
  • Deploy your RS (if you already have RS apps, you need to UMA-enable them) – there are a couple of different approaches, gateway/proxy (could be in the cloud) and an SDK for UMA-enabling apps. The former seems preferred. Eve calls these "protectlets" (smile)
  • Encourage data partners to implement their own RS's – if you have a wide data ecosystem (i.e., multiple RS's in different domains), your partners will have to UMA-enable their hosts. Helping them do this is good business! See just above. Nancy notes that Patient Centric Solutions' new PatientShare offering does exactly this.

This whole list opens up the idea of publishing an auxiliary document called something like an UMA Deployment Guide and stimulating open source (once more) to encourage deployments.

Proposal: Lisa suggests that, since the Operators are extremely likely to be Legal Persons, we should acknowledge the obvious and enhance the formal definitions in the document somehow so that it won't trip up people who read them there. Maybe we can insert some bracketed text and/or italicize some? She will draft some candidate text and let's discuss it on the list. There are so many considerations around this question – rhetorical, UMA maturity, etc., that it's not even funny.

Domenico comments on the concept of a "data ecosystem". We've been talking about wide ecosystems for a while, but it seems there are different kinds of wide ecosystems. Lisa says: "UMA is an ecosystem builder." Nice! Eve has often called OAuth2 a protocol framework because it has so many choices that it's not just one protocol; unfortunately that has risked interoperability and security. XYZ is trying to enable protocol-building without all those sacrifices. UMA is meant to make it possible to build data – and data-sharing – ecosystems without sacrifices and compromises.

2019-11-05

Attending: Eve, Nancy, Tim (regrets: Cigdem, Andi)

...